research article open access drugs affecting the renin ...drugs affecting the renin-angiotensin...

Cardwell et al. BMC Medicine 2014, 12:28http://www.biomedcentral.com/1741-7015/12/28

RESEARCH ARTICLE Open Access

Drugs affecting the renin-angiotensin system andsurvival from cancer: a population based study ofbreast, colorectal and prostate cancer patientcohortsChris R Cardwell1*, Úna C Mc Menamin1, Blánaid M Hicks1, Carmel Hughes2, Marie M Cantwell1

and Liam J Murray1

Abstract

Background: Angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin II receptor blockers (ARBs) arecommonly prescribed to the growing number of cancer patients (more than two million in the UK alone) often totreat hypertension. However, increased fatal cancer in ARB users in a randomized trial and increased breast cancerrecurrence rates in ACEI users in a recent observational study have raised concerns about their safety in cancerpatients. We investigated whether ACEI or ARB use after breast, colorectal or prostate cancer diagnosis wasassociated with increased risk of cancer-specific mortality.

Methods: Population-based cohorts of 9,814 breast, 4,762 colorectal and 6,339 prostate cancer patients newlydiagnosed from 1998 to 2006 were identified in the UK Clinical Practice Research Datalink and confirmed by cancerregistry linkage. Cancer-specific and all-cause mortality were identified from Office of National Statistics mortalitydata in 2011 (allowing up to 13 years of follow-up). A nested case–control analysis was conducted to compareACEI/ARB use (from general practitioner prescription records) in cancer patients dying from cancer with up to fivecontrols (not dying from cancer). Conditional logistic regression estimated the risk of cancer-specific, and all-cause,death in ACEI/ARB users compared with non-users.

Results: The main analysis included 1,435 breast, 1,511 colorectal and 1,184 prostate cancer-specific deaths(and 7,106 breast, 7,291 colorectal and 5,849 prostate cancer controls). There was no increase in cancer-specificmortality in patients using ARBs after diagnosis of breast (adjusted odds ratio (OR) = 1.06 95% confidence interval (CI)0.84, 1.35), colorectal (adjusted OR = 0.82 95% CI 0.64, 1.07) or prostate cancer (adjusted OR = 0.79 95% CI 0.61, 1.03).There was also no evidence of increases in cancer-specific mortality with ACEI use for breast (adjusted OR = 1.06 95% CI0.89, 1.27), colorectal (adjusted OR = 0.78 95% CI 0.66, 0.92) or prostate cancer (adjusted OR = 0.78 95% CI 0.66, 0.92).

Conclusions: Overall, we found no evidence of increased risks of cancer-specific mortality in breast, colorectal orprostate cancer patients who used ACEI or ARBs after diagnosis. These results provide some reassurance thatthese medications are safe in patients diagnosed with these cancers.

Keywords: Colorectal cancer, Breast cancer, Prostate cancer, Mortality, Angiotensin-converting enzymeinhibitors and angiotensin II receptor blockers

* Correspondence: [email protected] Epidemiology and Health Services Research Group, Centre for PublicHealth, Queen’s University Belfast, Belfast, Northern IrelandFull list of author information is available at the end of the article

© 2014 Cardwell et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly credited.

mailto:[email protected]

http://creativecommons.org/licenses/by/2.0

Cardwell et al. BMC Medicine 2014, 12:28 of 15http://www.biomedcentral.com/1741-7015/12/28

BackgroundAngiotensin-converting enzyme inhibitors (ACEIs) andangiotensin II receptor blockers (ARBs) are commonlyprescribed to the large and growing number of individualswith cancer (for example, currently more than two millionin the UK [1] and 13 million in the US [2]) often to treathypertension which affects around 40% of cancer patients[3]. However, the possible effect of ACEIs and ARBs oncancer is subject to much debate. Concerns were first raisedin 2003 when the Candesartan in Heart Failure Assessmentof Reduction in Mortality and Morbidity (CHARM) trial[4] unexpectedly reported a significant 40% increase in fatalcancer in patients randomized to candesartan (an ARB)compared with placebo (relative risk (RR) = 1.42 95% CI1.02, 1.98; P = 0.04). In part conducted because of this find-ing, a 2010 meta-analysis of randomized trials observed in-creased cancer risk with ARBs [5]. Preclinical studies alsosuggested a biological rationale for an increase in cancerrisk in ARB users [6,7]. A later meta-analysis observed noassociation for ARBs but observed a small increased riskof cancer in combination ARB and ACEI users [8]. Thesefindings for cancer risk were further investigated in largepopulation-based observational studies [9-12] which,although generally negative, observed small increases incancer risk, for instance, for breast and prostate cancer inARB users [12] and lung cancer in ACEI users [11].These results, and particularly the increases in fatal

cancer observed in the 2003 ARB trial [4], raise ques-tions about the safety of ACEI and ARB use in cancerpatients. Despite these findings, there have not been anystudies which have investigated ARB use separately aftercancer diagnosis and cancer progression in prostate orcolorectal cancer patients and only two studies haveinvestigated the specific association between ARB useand cancer progression in breast cancer patients [13,14].ACEIs and cancer progression has also received little atten-tion with only two studies of ACEI use in colorectal cancerpatients [15,16] and one in prostate cancer patients [17].Five studies have investigated ACEI use in breast cancer pa-tients and cancer progression but have reached conflictingresults [13,14,18-20]. Further investigation of ACEIs andARBs and cancer progression is, therefore, warranted.This study investigated whether ACEI or ARB use after

diagnosis of breast, colorectal or prostate cancer was asso-ciated with increased cancer-specific, or all cause, mortalityin large population-based cohorts of cancer patients.

MethodsStudy designCohort studies were conducted utilizing linkages betweenthe English National Cancer Data Repository (NCDR), theUK Clinical Practice Research Datalink (CPRD) and theOffice of National Statistics (ONS) death registrations.The NCDR data include date and site of primary cancer

diagnosis and clinical data, such as stage and treatment.The CPRD is the world’s largest database of longitudinalpatient records comprising around 8% of the UK popula-tion and includes demographic information, clinical diag-noses and prescription data which is of documented highquality [21,22]. Ethical approval for all observationalresearch using CPRD data has been obtained from amulticenter research ethics committee [23]. Linkagesbetween the datasets were conducted using a deter-ministic algorithm based upon National Health Service(NHS) number, gender, date of birth and postcode. Patientswere included in the cohorts if they had a CPRD diag-nosis code for breast (women only), colorectal or prostatecancer which was confirmed by NCDR cancer diagnosis(based upon International Classification of Diseases (ICD)codes of C50 for breast cancer, C61 for prostate cancer,and C18, C19 and C20 for colorectal cancer) from 1998 to2006. Cancer patients with previous NCDR cancer diagno-sis, apart from in situ neoplasms and non-melanoma skincancers, were excluded. Cancer patients were also excludedif the date of cancer diagnosis preceded CPRD researchquality records. Date and cause of death up to 2011 weretaken from ONS. Analysis was restricted to individuals withavailable ONS mortality data from cancer diagnosis.

ACEI\ARB identificationACEIs and ARBs were defined as all agents within the twodrug classes according to the British National Formulary[24] (BNF, chapters 2.5.5.1 and 2.5.5.2, respectively). ACEIand ARB prescriptions within the cohorts from CPRDprescribing data were counted and converted to daily de-fined doses (DDD) on the basis of the quantity and strength(as defined by the World Health Organization [25]). Aquantity of 28 tablets was assumed for approximately 2% ofprescriptions where quantity was missing or inconsistent.Medication usage was ascertained in the exposure perioddescribed later.

Potential confoundersData available from the NCDR included stage, histologicalgrade, Gleason score (for prostate cancer), surgery, chemo-therapy and radiotherapy in the six months after diagnosis.Gleason score was converted to grade to increase com-pleteness [26]. General practitioner (GP) prescribing datawere used to determine hormone therapy in the first sixmonths after cancer diagnosis including androgen therapyfor prostate cancer (BNF chapter 8.3.4.2, including gonador-elin analogues and anti-androgens) and tamoxifen and aro-matase inhibitors for breast cancer (BNF chapter 8.3.4.1).Breast and prostate cancer patients were excluded ifhormone therapy preceded cancer diagnosis by eightweeks. In breast cancer patients, hormone replacementtherapy (HRT) for estrogen and progestogens (BNF chapters6.4.1. and 6.4.2.) was determined prior to diagnosis. Low


dose aspirin and statin use were taken from GP pre-scription records. Smoking, alcohol intake and bodymass index (BMI) were determined from the closest GPrecord prior to cancer diagnosis (records older than tenyears were ignored). Comorbidities were determined fromGP diagnosis codes on the basis of diagnoses contributingto a recent adaptation of the Charlson comorbidity indexfor GPRD [27].

Data analysisThe cancer cohorts were initially analyzed using a timematched nested case–control approach, a common ap-proach, for example [28], which accounts for immortal timebias [29] without requiring complicated statistical tech-niques [30] with minimal loss of precision [31], and a timevarying covariate approach, described later. Breast cancercases were members who had died due to breast cancer(with an ICD code of C50 as the underlying cause of death)and these were matched on age (in five year intervals) andyear of cancer diagnosis to five risk-set controls who livedat least as long after their cancer diagnosis. Correspond-ing analyses were conducted for colorectal cancer cases(ICD codes of C18, C19, C20, C21 or C26 as their under-lying cause of death) who were matched to risk-set controlson gender, site (colon or rectal), age (in five year intervals)and year of cancer diagnosis (in two year intervals) andprostate cancer cases (with ICD codes of C61 as theirunderlying cause of death) who were matched to riskset controls on age (in five year intervals) and year ofcancer diagnosis.The exposure period (for identification of post-diagnostic

medication usage) in cases was the period from cancerdiagnosis until six months prior to cancer-specific death.The exposure period in the controls was fixed to be thesame duration as that of their matched cases and beganat the date of cancer diagnosis in the control. The ex-posure period did not include prescriptions in the sixmonth period prior to death as these may reflect endof life treatment or increased exposure to healthcareprofessionals (sensitivity analyses investigated increasingthis to 12 months). Analyses were restricted to individualswith at least one year of follow-up after cancer diagnosis.Conditional logistic regression was used to estimate the

odds of death from cancer in cancer patients prescribedone or more ACEIs in the exposure period compared tothose with none and corresponding odds ratios (ORs) and95% confidence intervals (95% CIs) were determined beforeand after adjustment for potential confounders. Similaranalyses were conducted by duration of medication usage(investigating individuals prescribed over 365 DDDs ofACEIs in the exposure period equivalent to one year ofusage). These analyses were repeated for ARBs. An analysisof all-cause mortality was also conducted in which cancerpatients who died from any-cause were matched to risk-set

controls (using the same matching criteria as in the cancer-specific analyses) and conditional logistic regression modelswere applied as described previously. A secondary analysiswas conducted to investigate pre-diagnostic ACEI/ARBuse. Two sensitivity analyses were conducted which re-stricted the non-user group to patients with more similarindications. In one analysis, the cancer cohorts were firstrestricted to individuals using any antihypertensive medica-tions in the year prior to cancer diagnosis (including di-uretics, vasodilator antihypertensive drugs, centrally actingantihypertensive drugs, alpha-adrenoceptor blocking drugs,beta-blockers, ACEIs, ARBs, renin inhibitors, and calciumchannel blockers) and the analysis was conducted as previ-ously. In the other sensitivity analysis, the full cohort wasused as in the main analysis but ORs were calculated com-paring ACEI/ARB users in the exposure period not toACEI/ARB non-users in the exposure period (as in themain analysis) but to ACEI/ARB non-users who had usedat least one other antihypertensive medication in the expos-ure period. Sensitivity analyses were also conducted add-itionally adjusting for other antihypertensive use and inbreast and prostate cancer patients, adjusting for hormonetherapy at any time after diagnosis (not just in the first sixmonths after diagnosis as in the main analysis). A sensitivityanalyses was conducted in prostate cancer patients adjust-ing for Gleason score and in breast cancer patients re-stricted to cancer registries with high rates of availablestage (overall, over 85% complete). Stratified analyses wereconducted by stage (for colorectal and breast cancer), sex(for colorectal) and site (for colorectal cancer). Stratifiedanalyses were conducted after re-matching cases to controlswithin strata. Interaction tests were used to compare associ-ations between strata [32].Additionally, the cancer cohort was analyzed, without

conversion to case–control data, applying survival ana-lysis to investigate ACEI/ARB exposure as a time vary-ing covariate (individuals were considered non-usersprior to use and users after a lag of six months after firstuse, to mimic the case–control analysis) [29]. A similardose response analysis was conducted with individualsconsidered non-users prior to six months after firstuse, a short term user between six months after firstuse and six months after 365 DDDs and a longer termuser after this time. A separate analysis was conductedadditionally adjusting for competing risk of deathsusing Fine and Gray’s proportional subhazards model(not shown, as results similar) [33].A pre-study power calculation based upon 15% ACEI

use and predicted numbers suggested the study wouldhave over 80% power to detect an OR of 1.20 for cancer-specific mortality in patients receiving ACEIs in each ofthe cancer cohorts (breast, colorectal, prostate). More ac-curately, based upon the final numbers (shown in Table 1)and ACEI and ARB use (shown in Table 2), the study had

Table 1 Characteristics of cancer patients who die from cancer (cases) compared with controls, by cancer site

Breast cancer Colorectal cancer Prostate cancer

Cases Controls P Cases Controls P Cases Controls P

number (%) number (%) number (%) number (%) number (%) number (%)

Year of diagnosis

1998 to 2000 500 (35) 2,466 (35) N.A. 379 (25) 1,842 (25) N.A. 396 (33) 1,941 (33) N.A.

2001 to 2003 518 (36) 2,570 (36) 564 (37) 2,721 (37) 461 (39) 2,281 (39)

2003 to 2006 417 (29) 2,070 (29) 568 (38) 2,728 (37) 327 (28) 1,627 (28)

Age at diagnosis (years)

<50 339 (24) 1,667 (23) N.A. 90 (6) 330 (5) N.A. 6 (1) 20 (0) N.A.

50 to 59 286 (20) 1,430 (20) 226 (15) 1,117 (15) 71 (6) 351(6)

60 to 69 273 (19) 1,365 (19) 401 (27) 2,001 (27) 280 (24) 1,400 (24)

70 to 79 318 (22) 1,588 (22) 496 (33) 2,472 (34) 532 (45) 2,660 (45)

≥ 80 219 (15) 1,056 (15) 298 (20) 1,371 (19) 295 (25) 1,418 (24)

Male 0 (0) 0 (0) N.A. 871 (58) 4,231 (58) N.A. 1,184(100) 5,849 (100) N.A.

Exposure period (years):

Mean (sd) 3.9 (2.3) 3.9 (2.3) N.A. 2.8 (1.6) 2.9 (1.7) N.A. 3.8 (2.2) 3.7 (2.2) N.A.

Range 1.0 to 12.8 1.0 to 12.8 N.A. 1.0 to 10.5 1.0 to 10.5 N.A. 1.0 to 11.9 1.0 to 11.9

Stage:1 72 (11) 1,471 (43) <0.001 61 (6) 1,044 (18) <0.001

2 402 (61) 1,676 (49) 275 (25) 2,496 (44)

3 116 (18) 220 (6) 558 (51) 1,981 (35)

4 64 (10) 40 (1) 203 (19) 195 (3)

Missing 781 3,699 414 1,575

Grade:Well 54 (6) 751 (19) 81 (7) 562 (9) <0.001 34 (4) 538 (13) <0.001

Moderate 377 (41) 1,900 (49) 889 (75) 4,894 (78) 217 (28) 2,067 (48)

Poor 492 (53) 1,217 (31) 208 (17) 818 (13) 516 (67) 1,683 (39)

Missing 512 3,280 333 1017 417 1561

Treatment within six monthsof diagnosis

Chemotherapy 573 (40) 1,611 (23) <0.001 651 (43) 1,916 (26) <0.001 49 (4) 137 (2) <0.001

Radiotherapy 700 (49) 3,355 (47) 0.22 302 (20) 1,066 (15) <0.001 246 (21) 1,276 (22) 0.48

Surgery 1,087 (76) 5,982 (84) <0.001 1,214 (80) 65,15 (89) <0.001

Tamoxifen therapy 709 (49) 4,610 (65) <0.001

Androgen therapy 976 (82) 3,450 (59) <0.001

Radical prostatectomya 20 (2) 331 (8) <0.001

Smoking prior to diagnosis 0.03 0.01 <0.001

Non-smoker 710 (60) 3,827 (64) 640 (51) 3,138 (52) 453 (47) 2,541 (52)

Ex-smoker 226 (19) 1,072 (18) 365 (29) 1,885 (31) 331 (34) 1,660 (34)

Current smoker 249 (21) 1,094 (18) 249 (20) 983 (16) 185 (19) 674 (14)

Missing 250 1,113 257 1285 215 974

BMI prior to diagnosis 0.01 0.66 0.05

Mean (sd) 26.6 (5.5) 26.2 (5.1) 26.5 (4.7) 26.6 (4.8) 26.4 (4.0) 26.1 (3.7)

Missing 386 1,633 412 1,810 302 1,362

Alcohol prior to diagnosis 0.58 0.47 0.37

Alcohol consumer 831 (80) 4,333 (81) 950 (87) 4,585 (86) 791 (89) 4,047 (90)

Missing 401 1,779 420 1,977 298 1366


Table 1 Characteristics of cancer patients who die from cancer (cases) compared with controls, by cancer site(Continued)

Comorbidity (prior to diagnosisor during follow-up time)

Cerebrovascular 89 (6) 383 (5) 0.53 117 (8) 493 (7) 0.19 124 (10) 509 (9) 0.07

C.P.D. 264 (18) 1,247 (18) 0.60 236 (16) 1295 (18) 0.06 229 (19) 1,110 (19) 0.80

C.H.D. 60 (4) 249 (4) 0.68 91 (6) 367 (5) 0.13 108 (9) 395 (7) 0.01

Diabetes 126 (9) 468 (7) 0.02 192 (13) 866 (12) 0.21 141 (12) 619 (11) 0.17

Myocardial infarction 26 (2) 148 (2) 0.74 104 (7) 454 (6) 0.32 130 (11) 522 (9) 0.03

Peptic ulcer disease 39 (3) 184 (3) 0.79 111 (7) 477 (7) 0.19 77 (7) 440 (8) 0.18

P.V.D. 51 (4) 165 (2) 0.003 64 (4) 331 (5) 0.70 113 (10) 396 (7) 0.001

Renal disease 78 (5) 382 (5) 0.92 107 (7) 478 (7) 0.45 138 (12) 509 (9) 0.001

Rheumatological 76 (5) 330 (5) 0.35 55 (4) 272 (4) 0.82 37 (3) 287 (5) 0.01aExcludes cancer registries without available data. C.P.D., chronic pulmonary disease; C.H.D., congestive heart disease; N.A., not applicable; P.V.D., peripheralvascular disease; sd, standard deviation.


approximately 80% power to detect, at the 5% significancelevel, an OR for cancer-specific mortality in ACEI usersof 1.25 and for ARB users of 1.35 in each of the cancercohorts. Statistical analyses were conducted in STATA11 (StataCorp, College Station, Texas).

ResultsPatient cohortsFigure 1 shows the selection of the final cohorts. Theaverage follow-up (in those not dying) was six years(range one to thirteen years) in each of the three cohorts.The crude rate of cancer-specific death was 28.0 per 1,000person years in the breast cancer cohort (based upon 1,440breast cancer-specific deaths in 51,507 person years), 78.5per 1,000 person years in the colorectal cancer cohort(based upon 1,528 colorectal cancer-specific deaths in19,462 person years), and 41.2 per 1,000 person years in theprostate cancer cohort (based upon 1,194 prostate cancer-specific deaths in 28,970 person years). These cohorts wereconverted to case–control data for the cancer-specific mor-tality analysis with 1,435 breast, 1,511 colorectal and 1,184prostate cases (cancer-specific deaths) and 7,106 breast,7,291 colorectal and 5,849 prostate risk-set controls.

Patient characteristicsTable 1 shows characteristics of cancer-specific deaths(cases) and controls. The average time to death and, hence,the end of the exposure period was 3.9, 2.8 and 3.8 yearsfor breast, colorectal and prostate cancer, respectively.Cancer-specific deaths (cases) had higher stage, higherhistological grade, more chemotherapy and less sur-gery. A higher proportion of cases had radiotherapy forcolorectal but not breast or prostate cancer. Hormonetherapy was less frequent in breast cancer cases butmore frequent in prostate cancer cases. Smoking rates wereslightly higher in cases. Rates of comorbidities, alcohol

consumption and BMI levels prior to diagnosis weregenerally similar between cases and controls (Table 1).

ACEI/ARB use and mortality in breast cancer patientsIn breast cancer patients (Table 2), the ACEI prescriptionsafter diagnosis were similar in those dying from cancer(cases) compared with controls (16.9% versus 16.4%,respectively; OR = 1.04, 95% CI 0.88, 1.22). Adjustmentfor potential confounders did not alter this OR markedly(OR = 1.06, 95% CI 0.89, 1.27). Similarly, no associationswere observed between cancer-specific mortality and ARBuse (adjusted OR = 1.06, 95% CI 0.84, 1.35). These findingswere little altered after adjustment for stage and were fairlyconsistent across sensitivity analyses (shown in Table 3).No associations were observed with pre-diagnosticuse (Table 3). Finally, analyses of post-diagnostic ACEI(adjusted OR = 1.04, 95% CI 0.92, 1.19) and ARB (adjustedOR = 0.96, 95% CI 0.81, 1.14) use and all-cause mortalityshowed little evidence of associations (Table 4).

ACEI/ARB use and mortality in colorectal cancer patientsIn colorectal cancer patients (Table 2), there was someevidence of a reduction in cancer-specific mortality inpatients prescribed an ACEI (OR = 0.79, 95% CI 0.69, 0.92)which was little altered after adjustment for confounders(adjusted OR = 0.78, 95% CI 0.66, 0.92), but a clear doseresponse association was not apparent as there was littleevidence of protective effects in those using more than365 DDDs of ACEI (adjusted OR = 0.84 95% CI 0.68, 1.03).There was little evidence of an association between ARBsand cancer-specific mortality (adjusted OR = 0.82 95% CI0.64, 1.07). Additional adjustment for stage little alteredthese estimates. Sensitivity analyses (Table 5) producedfairly consistent results for ARBs. Any protective associ-ation for ACEIs were attenuated when prescriptions in theyear prior to death were removed (adjusted OR = 0.83 95%

Table 2 Association between post-diagnostic exposure to ACEIs and ARBS and cancer specific death, by cancer site

Post-diagnosticmedication usage

All cancer patients Cancer patients with available stage/gradec

Cancer specificdeaths number (%)

Controlsnumber (%)

UnadjustedOR (95% CI)

P (trenda) Adjustedb

OR (95% CI)P (trenda) Unadjusted

OR (95% CI)P (trenda) Fully adjusted

OR (95% CI)dP (trenda)

Breast cancer

ACEI

0 (non-user) 1,192 (83.1) 5,938 (83.6) 1.00 (ref. cat.) 1.00 (ref. cat.) 1.00 (ref. cat.) 1.00 (ref. cat.)

≥1 DDD (user) 243 (16.9) 1,168 (16.4) 1.04 (0.88, 1.22) 0.65 1.06 (0.89, 1.27) 0.52 0.96 (0.76, 1.21) 0.72 0.83 (0.63, 1.09) 0.18

(0.50) (0.32) (0.88) (0.47)

1to 365 DDDs 92 (6.4) 469 (6.6) 0.97 (0.77, 1.23) 0.80 0.96 (0.75, 1.24) 0.78 0.88 (0.61, 1.27) 0.49 0.65 (0.43, 0.99) 0.04

≥365 DDDs 151 (10.5) 699 (9.8) 1.08 (0.89, 1.32) 0.42 1.14 (0.92, 1.41) 0.25 1.01 (0.76, 1.33) 0.96 0.96 (0.69, 1.34) 0.82

ARB


≥1 DDD (user) 102 (7.1) 504 (7.1) 1.00 (0.80, 1.26) 0.98 1.06 (0.84, 1.35) 0.62 1.03 (0.84, 1.25) 0.80 0.94 (0.65, 1.37) 0.75

(0.99) (0.57) (0.69) (0.85)

1 to 365 DDDs 33 (2.3) 160 (2.3) 1.02 (0.70, 1.51) 0.87 1.01 (0.68, 1.51) 0.96 1.04 (0.58, 1.87) 0.90 0.84 (0.44, 1.61) 0.59

≥365 DDDs 69 (4.8) 344 (4.8) 0.99 (0.75, 1.30) 0.95 1.09 (0.82, 1.45) 0.55 1.08 (0.73, 1.60) 0.70 0.99 (0.64, 1.52) 0.95

Colorectal cancer

ACEI


≥1 DDD (user) 280 (18.5) 1,631 (22.4) 0.79 (0.69, 0.92) 0.002 0.78 (0.66, 0.92) 0.003 0.75 (0.63, 0.89) 0.001 0.76 (0.62, 0.93) 0.01

(0.01) (0.02) (0.005) (0.02)

1 to 365 DDDs 113 (7.5) 729 (10.0) 0.72 (0.59, 0.89) 0.002 0.71 (0.57, 0.89) 0.03 0.69 (0.54, 0.90) 0.005 0.71 (0.53, 0.94) 0.02

≥365 DDDs 167 (11.1) 902 (12.4) 0.85 (0.71, 1.02) 0.09 0.84 (0.68, 1.03) 0.10 0.80 (0.64, 0.98) 0.03 0.80 (0.63, 1.02) 0.07

ARB


≥1 DDD (user) 83 (5.5) 480 (6.6) 0.84 (0.66, 1.07) 0.15 0.82 (0.64, 1.07) 0.14 0.91 (0.76, 1.08) 0.29 0.80 (0.59, 1.09) 0.16

(0.41) (0.94) (0.26) (0.66)

1 to 365 DDDs 29 (1.9) 221 (3.0) 0.64 (0.43, 0.95) 0.03 0.63 (0.42, 0.94) 0.03 0.51 (0.31, 0.83) 0.01 0.61 (0.37, 1.03) 0.06

≥365 DDDs 54 (3.6) 259 (3.6) 1.00 (0.74, 1.36) 0.98 1.00 (0.72, 1.37) 0.98 0.98 (0.69, 1.38) 0.89 0.93 (0.64, 1.36) 0.72

Prostate cancer

ACEI

0 (non-user) 848 (71.6) 4,145 (70.9) 1.00 (ref. cat.) 1.00 (ref. cat.) 1.00 (ref. cat.) 1.00 (ref. cat.)

≥1 DDD (user) 336 (28.4) 1,704 (29.1) 0.96 (0.83, 1.10) 0.55 0.78 (0.66, 0.92) 0.003 0.98 (0.82, 1.17) 0.80 0.82 (0.67, 1.02) 0.07

(0.47) (0.003) (0.85) (0.11)

Cardw

elletal.BM

CMedicine

2014,12:28Page

6of

15http://w

ww.biom

edcentral.com/1741-7015/12/28

Table 2 Association between post-diagnostic exposure to ACEIs and ARBS and cancer specific death, by cancer site (Continued)

1 to 365 DDDs 118 (10.0) 572 (9.8) 1.00 (0.81, 1.24) 0.99 0.83 (0.66, 1.04) 0.11 0.96 (0.74, 1.26) 0.79 0.80 (0.59, 1.07) 0.14

≥365 DDDs 218 (18.4) 1,132 (19.4) 0.94 (0.79, 1.11) 0.44 0.75 (0.62, 0.91) 0.003 0.98 (0.80, 1.21) 0.88 0.84 (0.66, 1.07) 0.16

ARB


≥1 DDD (user) 81 (6.8) 465 (8.0) 0.86 (0.67, 1.10) 0.22 0.79 (0.61,1.03) 0.08 0.93 (0.70, 1.23) 0.62 0.82 (0.61, 1.11) 0.21

(0.16) (0.06) (0.75) (0.26)

1 to 365 DDDs 30 (2.5) 153 (2.6) 0.97 (0.65, 1.44) 0.88 0.88 (0.58, 1.33) 0.55 0.84 (0.52, 1.36) 0.48 0.77 (0.46, 1.29) 0.33

≥365 DDDs 51 (4.3) 312 (5.3) 0.80 (0.59, 1.09) 0.15 0.74 (0.54, 1.02) 0.07 0.98 (0.70, 1.36) 0.90 0.84 (0.59, 1.21) 0.36aTests for trend from conditional logistic regression model based upon categories of DDDs (0, 1 to 365, ≥365 DDDs coded as 0, 1, 2, respectively). bAll models include surgery (within six months of diagnosis),chemotherapy (within six months), radiotherapy (within six months), low dose aspirin (during exposure period), statins (during exposure period), comorbidities (pre-diagnosis or during exposure period, includingmyocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, renal disease, peptic ulcer disease and diabetes), and smoking (pre-diagnosis, withmissing included as a category). The breast cancer model additionally includes tamoxifen (within six months), aromatase inhibitors (within six months), hormone replacement therapy (pre-diagnosis). The prostatecancer model additionally includes androgen therapy (within six months). cAnalysis includes 1,093 cases and 5,231 controls in colorectal cancer patients with available stage, 648 cases and 3,193 controls in breastcancer patients with available stage and 766 cases and 3,777 controls in prostate cancer patients with available grade. dAdditionally adjusted for stage in colorectal cancer patients and breast cancer patients andadjusted for grade in prostate cancer patients. ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CI, confidence interval; DDDs, daily defined doses; OR, odds ratio.

Cardw

elletal.BM

CMedicine

2014,12:28Page

7of

15http://w

ww.biom


1,435 cancer-specific deaths (cases) 1,511 cancer-specific deaths 1,184 cancer-specific deaths7,106 matched controls 7,291 matched controls 5,849 matched controls(5 cases had no available match) (17 cases had no available match) (10 cases had no available match)

2,258 all-cause deaths 2,101 all-cause deaths 2,197 all-cause deaths11,154 matched controls 10,107 matched controls 10,838 matched controls(13 cases had no available match) (33 cases had no available match) (16 cases no available match)

Patients diagnosed 1998-2007 with CPRD coverage and no previous cancer.

Breast cancer Colorectal cancer Prostate cancer

11,863 breast cancer patients 7,081 colorectal cancer patients 8,128 prostate cancer patients

After excluding patients where cancer diagnosis preceded CPRD research quality records.*


After excluding patients where death registration data unavailable.† 10,853 breast cancer patients 6,540 colorectal cancer patients 7,388 prostate cancer patients

After excluding patients with <1 year follow-up. 9,971 breast cancer patients 4,762 colorectal cancer patients 6,513 prostate cancer patients

After excluding patients where hormone therapy preceded cancer diagnosis by 8 weeks.‡


Final cohort for analysis. 9,814 breast cancer patients 4,762 colorectal cancer patients 6,339 prostate cancer patients 1,440 cancer specific deaths 1,528 cancer specific deaths 1,194 cancer specific deaths

Data for analysis after conversion to case-control data.

2,271 all-cause deaths 2,134 all-cause deaths 2,213 all-cause deaths

Cancer-specific mortality analysis

All-cause mortality analysis

* Patients excluded because cancer diagnosis preceded CPRD research quality records, and therefore prescription records may not be complete from diagnosis. † Patients excluded because death registration data unavailable and therefore death and cause of death data would not be identified in this patients. ‡ Patients excluded because hormone therapy records preceding cancer diagnosis suggests an incorrect diagnosis date, not applicable to colorectal cancer.

Figure 1 Flow chart of selection of patients for inclusion in analysis, by cancer site.


CI 0.67, 1.03) or when the analysis was restricted to usersof any antihypertensive medication in the year prior tocancer diagnosis (adjusted OR = 0.83 95% CI 0.65, 1.07).In early stage colorectal cancer a more marked reductionin cancer specific mortality was observed with ACEIuse (adjusted OR = 0.54 95% CI 0.38, 0.77) comparedwith later stage disease (adjusted OR = 0.86, 95% CI0.68, 1.10; P for interaction = 0.03). In general, any pro-tective associations were less marked when using thetime varying covariate approach but other sensitivityanalyses for ACEIs were similar to the main finding.No associations were observed between cancer-specificmortality and pre-diagnostic ACEI and ARB use (Table 3).Finally, there was little evidence of associations withall-cause mortality and ACEI (adjusted OR = 0.90, 95% CI0.79, 1.02) or ARB (adjusted OR = 0.94, 95% CI 0.78, 1.13)use after cancer diagnosis (Table 4).

ACEI/ARB use and mortality in prostate cancer patientsIn prostate cancer patients (Table 2), there was no evidenceof an increase in cancer-specific mortality in ACEI users(adjusted OR = 0.78, 95% CI 0.66, 0.92) or ARB users(adjusted OR= 0.79, 95% CI 0.61, 1.03). Additionally, adjust-ing for grade little altered the point estimates although the

protective association with ACEIs was no longer significant(fully adjusted OR = 0.82 95% CI 0.67, 1.02). The associa-tions for ARB use were fairly consistent across sensitiv-ity analyses (Table 6). The protective association withACEIs was more marked when restricting the analysisto antihypertensive users during the exposure period(adjusted OR = 0.78 95% CI 0.63, 0.96) and when adjustingfor other antihypertensive use (adjusted OR = 0.74 95% CI0.60, 0.92), but were attenuated in most other sensitivityanalyses particularly when prescriptions in the year prior todeath were removed (adjusted OR = 0.88 95% CI 0.70, 1.09)and in time-varying covariate analyses (adjusted hazardratio (HR) = 0.90 95% CI 0.76, 1.07). There was littleevidence of a reduction in all-cause mortality in ACEIusers (adjusted OR = 0.91, 95% CI 0.82, 1.03) or ARBusers (adjusted OR = 0.85, 95% CI 0.72, 1.01) after cancerdiagnosis (Table 4).

DiscussionOverall, we found no evidence of increased risks ofcancer-specific or all-cause mortality in breast, colorectalor prostate cancer patients using ACEIs or ARBs aftercancer diagnosis. There was some evidence of reductionsin the risk of cancer-specific mortality in colorectal and

Table 3 Sensitivity analyses for association between ACEIs and ARBS and cancer specific mortality in breast cancer patients

Cancer specificdeaths number

Controlsnumber

OR (95% CI) P OR (95% CI) P

ACEI user versusnon user

ARB user versusnon user

Breast cancer

Main analysis: diagnosis to six months prior to deatha 648 3,193 0.83 (0.63, 1.09) 0.18 0.94 (0.65, 1.37) 0.75

Diagnosis to 1 year prior to deathb 583 2,875 0.87 (0.65, 1.16) 0.33 0.87 (0.58, 1.30) 0.49

Restricted to users of any antihypertensive medicationc priorto cancer diagnosisd

212 994 0.87 (0.60, 1.25) 0.44 1.36 (0.85, 2.16) 0.20

Comparison group restricted to users of any antihypertensivein exposure periodc

361 1,624 0.79 (0.59, 1.05) 0.10 0.92 (0.63, 1.34) 0.67

Additionally adjusting for other antihypertensivese 648 3,193 0.79 (0.60, 1.06) 0.12 0.90 (0.61, 1.33) 0.60

Additionally adjusting for hormone therapy any time after diagnosisf 648 3,193 0.80 (0.60, 1.05) 0.11 0.98 (0.66, 1.44) 0.91

≥730 DDDs versus 0 DDDs (non-user) 648 3,193 1.06 (0.74, 1.52) 0.75 0.99 (0.64, 1.52) 0.95

Restricted to cancer registries with stage data available for over85% of patients

487 1,911 0.84 (0.59, 1.18) 0.31 1.33 (0.81, 2.17) 0.25

Stage 1 and 2g 469 2,313 0.95 (0.68, 1.32) 0.75 1.14 (0.73, 1.81) 0.56

Stage 3 and 4g 161 531 0.83 (0.46, 1.51) 0.54 2.37 (1.05, 5.35) 0.04

Pre-diagnostic useh 695 3,413 0.86 (0.61, 1.20) 0.37 1.06 (0.65, 1.72) 0.82

Time varying covariate analysisi 656 4,164

≥1 DDDs versus 0 DDDs (non-user) 0.97 (0.77, 1.21) 0.76 1.20 (0.89, 1.62) 0.23

1to 365 DDDs versus 0 DDDs (non-user) 0.90 (0.65, 1.25) 0.52 0.94 (0.55, 1.60) 0.81

≥365 DDDs versus 0 DDDs (non-user) 1.01 (0.77, 1.32) 0.93 1.35 (0.96, 1.91) 0.09aExcept where otherwise stated analysis investigates medications from diagnosis to six months prior to death/index date and models include surgery (within sixmonths of diagnosis), chemotherapy (within six months), radiotherapy (within six months), low dose aspirin (during exposure period), statins (during exposure period),comorbidities (pre-diagnosis or during exposure period, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease,peripheral vascular disease, renal disease, peptic ulcer disease and diabetes), and smoking (pre-diagnosis, with missing included as a category), tamoxifen (within six months),aromatase inhibitors (within six months), hormone replacement therapy (pre-diagnosis) and stage. bRestricted to individuals surviving more than 1.5 years. cAntihypertensivemedications include beta-blockers, diuretics, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, alpha-adrenoceptor blocking drugs, ACEIs, ARBs, renininhibitors, and calcium channel blockers. dRestricted to individuals with >1 year of records prior to cancer diagnosis and pre-diagnostic use considered antihypertensiveuse in that year, excluding deaths in the year after cancer diagnosis. eModels include all variables in aand additionally include calcium channel blockers, diuretics andbeta-blockers. fModels include all variables in abut tamoxifen and aromatase inhibitors were determined at any time after diagnosis in the exposure period. gStratifiedanalyses were conducted after re-matching controls to cases within strata; due to non-availability of matches overall numbers in subgroups may not be identical tonumbers presented in Table 1. hRestricted to individuals with >1 year of records prior to cancer diagnosis and pre-diagnostic use considered ACEI/ARB use in that year,not excluding deaths in the year after cancer diagnosis. iReported estimates are adjusted hazard ratios and 95% CIs, model includes age at diagnosis, year of diagnosis,surgery (within six months of diagnosis), chemotherapy (within six months), radiotherapy (within six months), statins (during exposure period, as a time varyingcovariate), comorbidities (pre-diagnosis including, myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheralvascular disease, renal disease, peptic ulcer disease and diabetes) and stage. ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers;CI, confidence interval; DDDs, daily defined doses; OR, odds ratio.


prostate cancer patients using ACEIs, but any protectiveeffects were weak in magnitude, were inconsistent acrosssensitivity analyses and were not a priori stated and,therefore, are difficult to interpret.

Strengths and weaknessesThe main strengths of our study were the large size(including over 20,000 cancer patients and 4,000 cancer-specific deaths) and the long duration of follow-up(up to 13 years) which provided the ability to detectrelatively weak effects and to report narrow CIs whichrule out relatively small increases in risk. Our studyused GP prescribed drug information which capturesalmost all ACEI and ARB use (which are only available byprescription in the UK), eliminates the potential for recallbias incurred by self–report and allows detailed temporal

associations to be explored. Consequently, our data reflectGP prescriptions rather than drug consumption butanalyses of multiple prescriptions generally found similarresults, suggesting compliance may not impact our resultsgreatly. As with all observational studies we cannot excluderesidual or unknown confounding. We had robust dataon surgery, radiotherapy, chemotherapy, hormone ther-apy, comorbidities and importantly stage for colorectaland breast and Gleason score for prostate cancer, butwe had limited data on smoking and alcohol intake andno information on socioeconomic status.

Comparison with previous studiesIn prostate and colorectal cancer our study is the firstepidemiological study to investigate ARBs separately andcancer progression and only three studies have investigated

Table 4 Association between post-diagnostic exposure to ACEIs and ARBS and all-cause mortality, by cancer site

Post-diagnosticmedication usage

All cancer patients Cancer patients with available stage/gradeb

All-cause deathsnumber (%)

Controlsnumber (%)

UnadjustedOR (95% CI)

P (trendd) Adjusteda

OR (95% CI)P (trendd) Unadjusted

OR (95% CI)P (trendd) Fully adjusted

OR (95% CI)cP (trendd)

Breast cancer

ACEI


≥1 DDD (user) 509 (22.5) 2,206 (19.8) 1.19 (1.06, 1.34) 0.003 1.04 (0.92, 1.19) 0.51 1.19 (1.00, 1.41) 0.05 0.93 (0.76, 1.14) 0.49

(0.01) (0.60) (0.09) (0.37)

1 to 365 DDDs 193 (8.6) 821 (8.6) 1.20 (1.02, 1.42) 0.03 1.06 (0.89, 1.27) 0.52 1.26 (0.97, 1.62) 0.08 1.00 (0.76, 1.34) 0.97

≥365 DDDs 316 (14.0) 1,385 (12.4) 1.18 (1.03, 1.36) 0.02 1.03 (0.89, 1.20) 0.68 1.15 (0.94, 1.41) 0.17 0.89 (0.71, 1.13) 0.34

ARB


≥1 DDD (user) 201 (8.9) 993 (8.9) 1.00 (0.85, 1.18) 0.99 0.96 (0.81, 1.14) 0.62 0.91 (0.71, 1.16) 0.45 0.79 (0.60 ,1.03) 0.08

(0.85) (0.80) (0.69) (0.18)

1 to 365 DDDs 64 (2.8) 338 (3.0) 0.93 (0.70, 1.22) 0.59 0.88 (0.66, 1.17) 0.37 0.73 (0.47, 1.15) 0.18 0.62 (0.39, 1.00) 0.05

≥365 DDDs 137 (6.1) 655 (5.9) 1.04 (0.85, 1.26) 0.71 1.00 (0.82, 1.23) 0.99 1.00 (0.75, 1.33) 0.99 0.87 (0.64, 1.19) 0.38

Colorectal cancer

ACEI


≥1 DDD (user) 505 (24.0) 2,529 (25.0) 0.96 (0.85, 1.07) 0.44 0.90 (0.79, 1.02) 0.10 0.94 (0.82, 1.07) 0.36 0.89 (0.76, 1.03) 0.12

(0.31) (0.05) (0.38) (0.10)

1 to 365 DDDs 206 (9.8) 985 (9.8) 1.00 (0.86, 1.18) 0.95 0.95 (0.80, 1.13) 0.58 0.94 (0.78, 1.14) 0.55 0.92 (0.75, 1.13) 0.41

≥365 DDDs 299 (14.2) 1,544 (15.3) 0.92 (0.80, 1.06) 0.27 0.85 (0.73, 1.00) 0.05 0.94 (0.80, 1.10) 0.43 0.86 (0.72, 1.04) 0.11

ARB


≥1 DDD (user) 162 (7.8) 813 (8.0) 0.97 (0.82, 1.16) 0.78 0.94 (0.78, 1.13) 0.50 1.01 (0.82, 1.23) 0.95 1.02 (0.82, 1.26) 0.88

(0.88) (0.86) (0.66) (0.72)

1 to 365 DDDs 59 (2.8) 344 (3.4) 0.84 (0.64, 1.12) 0.24 0.80 (0.60, 1.06) 0.12 0.85 (0.61, 1.19) 0.35 0.92 (0.65, 1.31) 0.66

≥365 DDDs 103 (4.9) 469 (4.6) 1.07 (0.86, 1.34) 0.54 1.05 (0.83, 1.32) 0.70 1.11 (0.86, 1.41) 0.42 1.07 (0.83, 1.39) 0.60

Prostate cancer

ACEI


≥1 DDD (user) 734 (33.4) 3,326 (30.7) 1.14 (1.03, 1.26) 0.01 0.91 (0.82, 1.03) 0.13 1.09 (0.97, 1.23) 0.15 0.88 (0.76, 1.00) 0.06

(0.04) (0.05) (0.13) (0.07)

Cardw

elletal.BM

CMedicine

2014,12:28Page

10of

15http://w

ww.biom


Table 4 Association between post-diagnostic exposure to ACEIs and ARBS and all-cause mortality, by cancer site (Continued)

1 to 365 DDDs 258 (11.7) 1,098 (10.1) 1.20 (1.04, 1.40) 0.01 1.00 (0.85, 1.17) 0.96 1.06 (0.89, 1.27) 0.53 0.88 (0.73, 1.07) 0.20

≥365 DDDs 476 (21.7) 2,228 (20.6) 1.11 (0.98, 1.24) 0.09 0.87 (0.76, 0.99) 0.04 1.11 (0.97, 1.27) 0.14 0.87 (0.75, 1.02) 0.09

ARB


≥1 DDD (user) 195 (8.9) 965 (8.9) 1.00 (0.85, 1.18) 0.98 0.85 (0.72, 1.01) 0.06 1.08 (0.89, 1.30) 0.45 0.92 (0.75, 1.12) 0.39

(0.99) (0.07) (0.33) (0.56)

1 to 365 DDDs 61 (2.8) 297 (2.7) 1.02 (0.77, 1.35) 0.89 0.86 (0.64, 1.15) 0.32 0.95 (0.68, 1.33) 0.77 0.81 (0.57, 1.14) 0.26

≥365 DDDs 134 (6.1) 668 (6.2) 0.99 (0.82, 1.21) 0.95 0.84 (0.69, 1.03) 0.10 1.14 (0.91, 1.43) 0.26 0.97 (0.76, 1.23) 0.81aAll models include surgery (within six months of diagnosis), chemotherapy (within six months), radiotherapy (within six months), low dose aspirin (during exposure period), statins (during exposure period),comorbidities (pre-diagnosis or during exposure period, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheral vascular disease, renal disease, pepticulcer disease and diabetes), and smoking (pre-diagnosis, with missing included as a category). The breast cancer model additionally includes tamoxifen (within six months), aromatase inhibitors (within six months),hormone replacement therapy (pre-diagnosis). The prostate cancer model additionally includes androgen therapy (within six months). bAnalysis includes 1,558 cases and 7,406 controls in colorectal cancer patientswith available stage, 1,008 cases and 4,900 controls in breast cancer patients with available stage and 1,520 cases and 7,473 controls in prostate cancer patients with available grade. cAdditionally adjusted for stage incolorectal cancer patients and breast cancer patients and adjusted for grade in prostate cancer patient. dTests for trend from conditional logistic regression model based upon categories of DDDs (0, 1 to365, ≥365DDDs coded as 0, 1, 2 respectively). ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CI, confidence interval; DDDs, daily defined doses; OR, odds ratio.

Cardw

elletal.BM

CMedicine

2014,12:28Page

11of

15http://w

ww.biom


Table 5 Sensitivity analyses for association between ACEIs and ARBS and cancer specific mortality in colorectalcancer patients


Controlsnumber

OR (95% CI) P OR (95% CI) P



Colorectal cancer

Main analysis: diagnosis to six months prior to deatha 1,093 5,231 0.76 (0.62, 0.93) 0.01 0.80 (0.59, 1.09) 0.16

Diagnosis to one year prior to deathb 869 4,152 0.83 (0.67, 1.03) 0.10 0.89 (0.64, 1.25) 0.51


427 1,934 0.83 (0.65, 1.07) 0.15 0.90 (0.63, 1.27) 0.54

Comparison group restricted to users of any antihypertensivein exposure periodc

596 1,845 0.75 (0.61, 0.92) 0.01 0.80 (0.59, 1.10) 0.16

Additionally adjusting for other antihypertensivese 1,093 5,231 0.75 (0.61, 0.92) 0.01 0.80 (0.59, 1.09) 0.16

≥730 DDDs versus 0 DDDs (non-user) 1,093 5,231 0.73 (0.62, 1.00) 0.03 0.93 (0.64, 1.36) 0.72

Stage 1 and 2f 328 1,483 0.54 (0.38, 0.77) 0.001 0.80 (0.47, 1.37) 0.41

Stage 3 and 4f 740 3,185 0.86 (0.68, 1.10) 0.23 1.05 (0.72, 1.55) 0.79

Colon 649 3,134 0.84 (0.64, 1.09) 0.19 0.77 (0.52, 1.15) 0.21

Rectal 444 2,097 0.67 (0.49, 0.92) 0.01 0.91 (0.56, 1.50) 0.72

Males 638 3,078 0.80 (0.62, 1.03) 0.08 0.93 (0.62, 1.41) 0.74

Females 455 2,153 0.70 (0.49, 0.98) 0.04 0.64 (0.40, 1.03) 0.06

Pre-diagnostic useg 1,641 7,815 1.06 (0.89, 1.27) 0.51 0.86 (0.62, 1.18) 0.35

Time varying covariate analysish 1,109 2,559


1 to365 DDDs versus 0 DDDs (non-user) 0.76 (0.59, 0.97) 0.03 0.64 (0.40, 1.03) 0.07


≥1 DDDs versus 0 DDDs (non-user) in stage 1 and 2 patients 0.62 (0.46, 0.84) 0.002 0.65 (0.41, 1.05) 0.08

≥1 DDDs versus 0 DDDs (non-user) in stage 3 and 4 patients 0.91 (0.74, 1.10) 0.33 0.97 (0.70, 1.34) 0.83aExcept where otherwise stated analysis investigates medications from diagnosis to six months prior to death/index date and models include surgery (within sixmonths of diagnosis), chemotherapy (within six months), radiotherapy (within six months), low dose aspirin (during exposure period), statins (during exposureperiod), comorbidities (pre-diagnosis or during exposure period, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronicpulmonary disease, peripheral vascular disease, renal disease, peptic ulcer disease and diabetes), smoking (pre-diagnosis, with missing included as a category) andstage. bRestricted to individuals surviving more than 1.5 years. cAntihypertensive medications include beta-blockers, diuretics, vasodilator antihypertensive drugs,centrally acting antihypertensive drugs, alpha-adrenoceptor blocking drugs, ACEIs, ARBs, renin inhibitors and calcium channel blockers. dRestricted to individualswith >1 year of records prior to cancer diagnosis and pre-diagnostic use considered antihypertensive use in that year, excluding deaths in the year after cancerdiagnosis. eModels include all variables in a and additionally include calcium channel blockers, diuretics and beta-blockers. fStratified analyses were conductedafter re-matching controls to cases within strata, due to non-availability of matches overall numbers in subgroups may not be identical to numbers presented inTable 1. gRestricted to individuals with >1 year of records prior to cancer diagnosis and pre-diagnostic use considered ACEI/ARB use in that year, not excludingdeaths in the year after cancer diagnosis. hReported estimates are adjusted hazard ratios and 95% CI, model contains age at diagnosis, year of diagnosis, gender,site (colon or rectum), surgery (within six months of diagnosis), chemotherapy (within six months), radiotherapy (within six months), statins (during exposureperiod as a time varying covariate), comorbidities (pre-diagnosis including myocardial infarction, cerebrovascular disease, congestive heart disease, chronicpulmonary disease, peripheral vascular disease, renal disease, peptic ulcer disease and diabetes) and stage. ACEIs, angiotensin-converting enzyme inhibitors; ARBs,angiotensin II receptor blockers; CI, confidence interval; DDDs, daily defined doses; OR, odds ratio.


ACEI and cancer progression, one [17] in prostate cancerpatients and two [15,16] in colorectal cancer patients.An earlier ACEI study in 62 prostate cancer patientsafter radical prostatectomy [17] observed a reduced rateof biochemical recurrence in ACEI users compared withnon-users (3/32 versus 10/30). An earlier ACEI study in55 stage 2 colorectal cancer patients [15] observed a re-duction in the risk of distant metastasis with frequentACEI use prior to diagnosis (adjusted OR = 0.22) and amore recent study in 262 stage 3 and 4 cancer patientsdid not present an estimate for ACEI or ARB use separatelybut did present a reduced risk of mortality in patients

simultaneously using beta-blockers and ACEIs or ARBs(HR = 0.50 95% CI 0.29, 0.85).Five independent epidemiological studies [13,14,18-20]

have previously investigated ACEI and cancer-specificmortality in breast cancer patients but only two havepreviously investigated ARB use separately. A recent largeDanish study [13] demonstrated a small but non-significantincrease in cancer recurrence with ACEI use but no as-sociation with ARB use (HR = 1.2 95% CI 0.97, 1.4 andHR = 1.1 95% CI 0.85, 1.3, respectively). Another recentUS study [14], from the same cohort as an earlier study[34], observed no association between either ACEI use

Table 6 Sensitivity analyses for association between ACEIs and ARBS and cancer specific mortality in prostatecancer patients


Controlsnumber

OR (95% CI) P OR (95% CI) P



Prostate cancer

Main analysis: diagnosis to six months prior to deatha 766 3,777 0.82 (0.67, 1.02) 0.07 0.82 (0.61, 1.11) 0.21

Diagnosis to one year prior to deathb 682 3,360 0.88 (0.70, 1.09) 0.24 0.89 (0.64, 1.23) 0.48

Adjusting for Gleason score along with all variables ina apart from grade 516 2,518 0.80 (0.62, 1.03) 0.09 0.89 (0.61, 1.31) 0.55


399 1,925 0.80 (0.62, 1.04) 0.09 0.89 (0.62, 1.26) 0.51

Comparison group restricted to users of any antihypertensive inexposure periodc

584 2,753 0.78 (0.63, 0.96) 0.02 0.80 (0.59, 1.08) 0.15

Additionally adjusting for other antihypertensivese 766 3,777 0.74 (0.60, 0.92) 0.01 0.75 (0.55, 1.02) 0.07

Additionally adjusting for hormone therapy any time after diagnosisf 766 3,777 0.82 (0.66, 1.01) 0.06 0.84 (0.62, 1.14) 0.27

≥730 DDDs versus 0 DDDs (non-user) 766 3,777 0.91 (0.70, 1.18) 0.49 0.84 (0.59, 1.21) 0.36

Pre-diagnostic useg 821 4,041 0.98 (0.78, 1.23) 0.87 1.14 (0.75, 1.75) 0.53

Time varying covariate analysish 772 3,838


1 to 365 DDDs versus 0 DDDs (non-user) 0.87 (0.70, 1.08) 0.21 0.88 (0.62, 1.26) 0.49

≥365 DDDs versus 0 DDDs (non-user) 0.94 (0.76, 1.18) 0.60 1.03 (0.72, 1.48) 0.86aExcept where otherwise stated analysis investigates medications from diagnosis to six months prior to death/index date and models include surgery (within sixmonths of diagnosis), chemotherapy (within six months), radiotherapy (within six months), low dose aspirin (during exposure period), statins (during exposureperiod), comorbidities (pre-diagnosis or during exposure period, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronicpulmonary disease, peripheral vascular disease, renal disease, peptic ulcer disease and diabetes), smoking (pre-diagnosis, with missing included as a category),androgen therapy (within six months) and grade. bRestricted to individuals surviving more than 1.5 years. cAntihypertensive medications include beta-blockers,diuretics, vasodilator antihypertensive drugs, centrally acting antihypertensive drugs, alpha-adrenoceptor blocking drugs, ACEIs, ARBs, renin inhibitors, and calciumchannel blockers. dRestricted to individuals with >1 year of records prior to cancer diagnosis and pre-diagnostic use considered antihypertensive use in that year,excluding deaths in the year after cancer diagnosis.eModels include all variables in aand additionally include calcium channel blockers, diuretics and beta-blockers. fModels include all variables in abut androgentherapy was determined at any time after diagnosis in the exposure period. gRestricted to individuals with >1 year of records prior to cancer diagnosis andpre-diagnostic use considered ACEI/ARB use in that year, not excluding deaths in the year after cancer diagnosis. hReported estimates are adjusted hazard ratiosand 95% CIs models contain year of diagnosis, age at diagnosis, surgery (within six months of diagnosis), chemotherapy (within six months), radiotherapy (withinsix months), comorbidities (pre-diagnosis, including myocardial infarction, cerebrovascular disease, congestive heart disease, chronic pulmonary disease, peripheralvascular disease, renal disease, peptic ulcer disease and diabetes), statin (as a time varying covariate) and grade. ACEIs, angiotensin-converting enzyme inhibitors;ARBs, angiotensin II receptor blockers; CI, confidence interval; DDDs, daily defined doses; OR, odds ratio.


(HR = 1.07 95% CI 0.65, 1.77) or ARB use (HR = 0.4195% CI 0.15, 1.13) and cancer-specific mortality. An earliersmaller study [18], including 174 breast cancer-specificdeaths in 1,779 breast cancer patients, observed no as-sociation with simultaneous ACEI and beta-blocker useand when investigating ACEI users exclusively observeda marked increase in cancer recurrence (HR = 1.56 95%CI 1.02, 2.39) but little evidence of an increase in breastcancer-specific mortality (HR = 1.27 95% CI 0.74, 2.19).One study observed little evidence of association forself-reported ACEI use (HR = 0.89 95% CI 0.60, 1.32)[20]. Finally another study observed a marked reductionin cancer recurrence in 168 stage 2 or 3 breast cancerpatients using ACEI or ARBs after diagnosis (HR = 0.5795% CI 0.37, 0.89, P =0.01) [19], but this estimate mayhave incurred some immortal time bias [29] as individualsusing ACEIs or ARBs after diagnosis were consideredusers from diagnosis in the analysis so in the period fromdiagnosis to medication use they could not have died.

Our study observed some protective effects of ACEIs,particularly in early stage colorectal cancer patients andprostate cancer patients, which, if real, would support thetheory that ACEIs have a role in cancer therapy [35,36].Components of the renin-angiotensin system (RAS) areexpressed in various cancer sites [37] and may contributeto processes important for cancer progression includingcell proliferation and apoptosis. In vitro and animal studieshave shown that both ACEIs and ARBs can suppress cellproliferation and tumor/metastasis growth in various can-cers including breast [38] and colorectal [39]. Angiogenesisalso appears to be an important process by which the RASsystem exerts pro-tumor effects and ACEIs and ARBs re-duce the expression of vascular endothelial growth factor(VEGF) and other angiogenic factors in both cell lines [38]and animal models [40]. However, the protective effects ob-served in colorectal and prostate cancer patients should beinterpreted cautiously because, as previously stated, the ef-fects were weak, inconsistent and not a priori stated.


ConclusionsIn conclusion, concerns about the safety of ACEIs andARBs in cancer patients have been raised by trial datashowing increases in fatal cancers in ARB users [4] andobservational data showing increases in breast cancerrecurrence rates in ACEI users [18]. In contrast, our studyprovides no evidence of increased risks of cancer-specificmortality in users of ACEIs or ARBs with breast, prostateor colorectal cancer and suggests that these medicationsare safe in patients diagnosed with these common cancers.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsStudy concept and design: CRC, CH, LJM. Acquisition of data: CRC, UCM, CH,MMC, LJM. Analysis and interpretation of data: all authors. Drafting of themanuscript: CRC. Critical revision of the manuscript for important intellectualcontent: all authors. Statistical analysis: CRC, UCM, BMH. Obtained funding:CRC, CH, LJM. All authors read and approved the final manuscript.

AcknowledgementsCRC was supported by a Health and Social Care Research and Development,Public Health Agency, Northern Ireland, funded UK NIHR awarded CareerDevelopment Fellowship. BMH and UCM were funded by Northern IrelandDepartment of Education and Learning PhD studentships. Access to theGPRD database was funded through the Medical Research Council’s licenseagreement with the UK Medicines and Healthcare Products RegulatoryAgency. However, the interpretation and conclusions contained in this studyare those of the authors alone. None of the funders had any role in the designand conduct of the study; collection, management, analysis, and interpretationof the data; and preparation, review, or approval of the manuscript.

Author details1Cancer Epidemiology and Health Services Research Group, Centre for PublicHealth, Queen’s University Belfast, Belfast, Northern Ireland. 2School ofPharmacy, Queen’s University Belfast, Belfast, Northern Ireland.

Received: 21 August 2013 Accepted: 14 January 2014Published: 13 February 2014

References1. Maddams J, Utley M, Moller H: Projections of cancer prevalence in the

United Kingdom, 2010–2040. Br J Cancer 2012, 107:1195–1202.2. de Moor JS, Mariotto AB, Parry C, Alfano CM, Padgett L, Kent EE, Forsythe L,

Scoppa S, Hachey M, Rowland JH: Cancer survivors in the United States:prevalence across the survivorship trajectory and implications for care.Cancer Epidemiol Biomarkers Prev 2013, 22:561–570.

3. Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL: Prognosticimportance of comorbidity in a hospital-based cancer registry.JAMA 2004, 291:2441–2447.

4. Pfeffer MA, Swedberg K, Granger CB, Held P, McMurray JJ, Michelson EL,Olofsson B, Ostergren J, Yusuf S, Pocock S: Effects of candesartan onmortality and morbidity in patients with chronic heart failure: theCHARM-Overall programme. Lancet 2003, 362:759–766.

5. Sipahi I, Debanne SM, Rowland DY, Simon DI, Fang JC: Angiotensin-receptorblockade and risk of cancer: meta-analysis of randomised controlled trials.Lancet Oncol 2010, 11:627–636.

6. Goldstein MR, Mascitelli L, Pezzetta F: Angiotensin-receptor blockade,cancer, and concerns. Lancet Oncol 2010, 11:817–818.

7. Clere N, Corre I, Faure S, Guihot AL, Vessieres E, Chalopin M, Morel A,Coqueret O, Hein L, Delneste Y, Paris F, Henrion D: Deficiency or blockadeof angiotensin II type 2 receptor delays tumorigenesis by inhibitingmalignant cell proliferation and angiogenesis. Int J Cancer 2010,127:2279–2291.

8. Bangalore S, Kumar S, Kjeldsen SE, Makani H, Grossman E, Wetterslev J,Gupta AK, Sever PS, Gluud C, Messerli FH: Antihypertensive drugs and risk

of cancer: network meta-analyses and trial sequential analyses of324,168 participants from randomised trials. Lancet Oncol 2011, 12:65–82.

9. Hallas J, Christensen R, Andersen M, Friis S, Bjerrum L: Long term use of drugsaffecting the renin-angiotensin system and the risk of cancer: a population-based case–control study. Br J Clin Pharmacol 2012, 74:180–188.

10. Pasternak B, Svanstrom H, Callreus T, Melbye M, Hviid A: Use of angiotensinreceptor blockers and the risk of cancer. Circulation 2011, 123:1729–1736.

11. Azoulay L, Assimes TL, Yin H, Bartels DB, Schiffrin EL, Suissa S: Long-termuse of angiotensin receptor blockers and the risk of cancer. PLoS One2012, 7:e50893.

12. Bhaskaran K, Douglas I, Evans S, van Staa T, Smeeth L: Angiotensin receptorblockers and risk of cancer: cohort study among people receivingantihypertensive drugs in UK General Practice Research Database.BMJ 2012, 344:e2697.

13. Sorensen GV, Ganz PA, Cole SW, Pedersen LA, Sorensen HT, Cronin-FentonDP, Game JP, Christiansen PM, Lash TL, Ahern TP: Use of β-blockers,angiotensin-converting enzyme inhibitors, angiotensin II receptorblockers, and risk of breast cancer recurrence: a Danish nationwideprospective cohort study. J Clin Oncol 2013, 31:2265–2272.

14. Chae YK, Brown EN, Lei X, Melhem-Bertrandt A, Giordano SH, Litton JK,Hortobagyi GN, Gonzalez-Angulo AM, Chavez-Macgregor M: Use of ACEinhibitors and angiotensin receptor blockers and primary breast canceroutcomes. J Cancer 2013, 4:549–556.

15. Heinzerling JH, Anthony T, Livingston EH, Huerta S: Predictors of distantmetastasis and mortality in patients with stage II colorectal cancer.Am Surg 2007, 73:230–238.

16. Engineer DR, Burney BO, Hayes TG, Garcia JM: Exposure to ACEI/ARB andbeta-blockers is associated with improved survival and decreased tumorprogression and hospitalizations in patients with advanced colon cancer.Transl Oncol 2013, 6:539–545.

17. Ronquist G, Frithz G, Wang YH, Lindeborg T: Captopril may reducebiochemical (prostate-specific antigen) failure following radicalprostatectomy for clinically localized prostate cancer. Scand J UrolNephrol 2009, 43:32–36.

18. Ganz PA, Habel LA, Weltzien EK, Caan BJ, Cole SW: Examining the influenceof beta blockers and ACE inhibitors on the risk for breast cancerrecurrence: results from the LACE cohort. Breast Cancer Res Treat 2011,129:549–556.

19. Chae YK, Valsecchi ME, Kim J, Bianchi AL, Khemasuwan D, Desai A, Tester W:Reduced risk of breast cancer recurrence in patients using ACEinhibitors, ARBs, and/or statins. Cancer Invest 2011, 29:585–593.

20. Holmes MD, Hankinson SE, Feskanich D, Chen WY: Beta blockers andangiotensin-converting enzyme inhibitors’ purported benefit on breastcancer survival may be explained by aspirin use. Breast Cancer Res Treat2013, 139:507–513.

21. Boggon R, van Staa TP, Chapman M, Gallagher AM, Hammad TA, RichardsMA: Cancer recording and mortality in the General Practice ResearchDatabase and linked cancer registries. Pharmacoepidemiol Drug Saf 2013,22:168–175.

22. Herrett E, Thomas SL, Schoonen WM, Smeeth L, Hall AJ: Validation andvalidity of diagnoses in the General Practice Research Database:a systematic review. Br J Clin Pharmacol 2010, 69:4–14.

23. Clinical Practice Research Datalink: Ethical review for CPRD protocols.[http://www.cprd.com/isac/otherinfo.asp]

24. British Medical Association and the Royal Pharmaceutical Society of GreatBritain: British National Formulary. [http://www.bnf.org/bnf/index.htm]

25. World Health Organisation: World Health Organisation CollaboratingCentre for Drug Statistics Methodology. [http://www.whocc.no/]

26. National Cancer Institute: Prostate cancer treatment. [http://www.cancer.gov/cancertopics/pdq/treatment/prostate/HealthProfessional/page2]

27. Khan NF, Perera R, Harper S, Rose PW: Adaptation and validation of theCharlson Index for Read/OXMIS coded databases. BMC Fam Pract 2010, 11:1.

28. Schade R, Andersohn F, Suissa S, Haverkamp W, Garbe E: Dopamineagonists and the risk of cardiac-valve regurgitation. N Engl J Med 2007,356:29–38.

29. Levesque LE, Hanley JA, Kezouh A, Suissa S: Problem of immortal time biasin cohort studies: example using statins for preventing progression ofdiabetes. BMJ 2010, 340:b5087.

30. Etminan M: Pharmacoepidemiology II: the nested case–control study–a novelapproach in pharmacoepidemiologic research. Pharmacotherapy 2004,24:1105–1109.

http://www.cprd.com/isac/otherinfo.asp

http://www.bnf.org/bnf/index.htm

http://www.whocc.no/

http://www.cancer.gov/cancertopics/pdq/treatment/prostate/HealthProfessional/page2

http://www.cancer.gov/cancertopics/pdq/treatment/prostate/HealthProfessional/page2


31. Essebag V, Platt RW, Abrahamowicz M, Pilote L: Comparison of nestedcase–control and survival analysis methodologies for analysis oftime-dependent exposure. BMC Med Res Methodol 2005, 5:5.

32. Altman DG, Bland JM: Interaction revisited: the difference between twoestimates. BMJ 2003, 326:219.

33. Fine JP, Gray RJ: A proportional hazards model for the subdistribution ofa competing risk. J Am Stat Assoc 1999, 94:496–509.

34. Melhem-Bertrandt A, Chavez-Macgregor M, Lei X, Brown EN, Lee RT,Meric-Bernstam F, Sood AK, Conzen SD, Hortobagyi GN, Gonzalez-Angulo AM:Beta-blocker use is associated with improved relapse-free survival inpatients with triple-negative breast cancer. J Clin Oncol 2011, 29:2645–2652.

35. Ager EI, Neo J, Christophi C: The renin-angiotensin system and malignancy.Carcinogenesis 2008, 29:1675–1684.

36. George AJ, Thomas WG, Hannan RD: The renin-angiotensin system andcancer: old dog, new tricks. Nat Rev Cancer 2010, 10:745–759.

37. Deshayes F, Nahmias C: Angiotensin receptors: a new role in cancer?Trends Endocrinol Metab 2005, 16:293–299.

38. Chen X, Meng Q, Zhao Y, Liu M, Li D, Yang Y, Sun L, Sui G, Cai L, Dong X:Angiotensin II type 1 receptor antagonists inhibit cell proliferation andangiogenesis in breast cancer. Cancer Lett 2013, 328:318–324.

39. Neo JH, Malcontenti-Wilson C, Muralidharan V, Christophi C: Effect of ACEinhibitors and angiotensin II receptor antagonists in a mouse model ofcolorectal cancer liver metastases. J Gastroenterol Hepatol 2007, 22:577–584.

40. Wang L, Cai SR, Zhang CH, He YL, Zhan WH, Wu H, Peng JJ: Effects ofangiotensin-converting enzyme inhibitors and angiotensin II type 1receptor blockers on lymphangiogenesis of gastric cancer in a nudemouse model. Chin Med J (Engl) 2008, 121:2167–2171.

doi:10.1186/1741-7015-12-28Cite this article as: Cardwell et al.: Drugs affecting the renin-angiotensinsystem and survival from cancer: a population based study of breast,colorectal and prostate cancer patient cohorts. BMC Medicine 2014 12:28.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit

research article open access drugs affecting the renin ...drugs affecting the renin-angiotensin...

Documents